The relationship between smoking during pregnancy and congenital malformations was studied in prospective studies of 33,434 births in the Northern California Kaiser-Permanente Birth Defects Study and 55,933 births in the Collaborative Perinatal Project (CPP). 28.4% of women were smokers in the Kaiser population and 47% smoked in the CPP population. The odds ratio for smoking during pregnancy and major malformations in Kaiser was 1.00 (95% C.I. 0.8-1.2) and the odds ratio for minor malformations was 0.9 (0.8-0.9) (pp.001). The relationship between smoking and 54 individual malformations was evaluated in the Kaiser population. Statistically significant postive associations were observed for ventral hernias (10.1 (1.1-91) and 'other major gut abnormalities' (12.6 (1.5-108)). However, for each malformation the estimates were based on only one unexposed case. Significant negative associations were found for ventricular septal defects (0.5 (0.2-1.0)), hydroceles (0.7 (0.6-0.9)), clubfoot (0.7 (0.6-0.9)), pigmented nevi (0.7 (0.6-0.9)), hemangiomas (0.8 (0.7-0.9)), and Down syndrome (0.2 (0.1-0.9)). To determine if the findings noted above were an artifact of multiple comparisons, 7 of these 8 malformations were tabulated by smoking status for women in the CPP. All but one of the associations were not confirmed in the CPP. The association between smoking and hemangiomas in the CPP was 0.8 (0.6-1.0) (p=0.03). The prevalence of hemangioma at birth was 3.9% in the Kaiser population and 0.5% in the CPP. Adjustment for ethnicity strengthened this association in both data sets. We conclude that smoking is unlikely to be responsible for an increase in malformations.